SAP unveils AI-focused developer tools at TechEd

Vector database added to SAP HANA Cloud

For AI data management, SAP is adding a vector database to SAP HANA Cloud database. It enables a wide variety of data sources, including unstructured formats and company data, according to Mueller.

The vector database stores and manages unstructured data, including text documents, images and audio and video, he said. This makes it possible to make associations and provide context around similar data sources, such as financial, supply chain, quality, logistics, HR, procurement and customer experience data.

"With the vector database we're getting closer to the Holy Grail of generative AI, where you can combine the strength of LLMs [large language models] with actual real-time company data," Mueller said. "This will lead to better generative AI results with fewer hallucinations. And we are helping developers to combine rich structured data in SAP systems with unstructured data and with LLMs."

The vector database extension to SAP HANA Cloud is available at no extra cost to customers.

Finally, SAP debuted AI Foundation, a collection of AI services designed to help developers create AI- and generative AI-based extensions and applications on SAP BTP, according to Mueller.

This includes ready-to-use AI services such as trained and retrainable models for functions like document processing, recommendations and machine translation, he said. It also includes access to top LLMs from SAP partner vendors,  including Aleph Alpha, Anthropic, Cohere and OpenAI, that customers can use in an SAP context.

"AI Foundation includes everything developers need to start creating business-ready AI tools on SAP BTP," Mueller said, "from ready-to-use AI services and access to the top large language models to vector database capabilities and AI runtime and lifecycle management."

SAP did not specify any availability dates or pricing for the new tools except that the vector database will be extended to SAP HANA Cloud at no extra cost to customers.

AI-based tools should improve developer productivity

While there has been an overload of announcements around AI in enterprise applications and tools, the addition of AI to SAP's developer tools is generally positive, said Joshua Greenbaum, principal at Enterprise Applications Consulting.

The addition of AI -- conventional and generative -- has contributed to better productivity and accuracy in software development, Greenbaum said.

"Specifically, what SAP is doing in adding this to Build Code is a big positive, as anything that helps developers get it right the first time can be very positive," he said. "The nice thing about using AI on the app [developer] side is that you don't have to worry that much about dirty data. This is fundamentally fairly ordinary stuff -- it compiles or it doesn't."

There's ambiguity about what's conventional AI and what's generative AI in these capabilities, but the generative AI will be useful in writing code for SAP developers, Greenbaum said.

"It's writing something for you. And as long as that code is carefully reviewed by humans, it's a good productivity tool," he said.

Tools that automatically generate code are not new. But SAP Build Code can help with the complexity of what SAP developers have to deal with around data, systems integration and resource management, Greenbaum said.

"With all these different components, having an intelligent agent sitting on your shoulder and pointing you in the right direction is good," he said.

However, developers that use AI must be careful when building specific line-of-business applications, Greenbaum said. For example, AI being used by companies to help create press releases for investor relations may include forward-looking information ahead of a legal reporting deadline.

"You have to be concerned that the [personally identifiable information] is being protected as you drive this functionality to improve customer or employee services," he said. "LLMs are essentially indiscriminate vacuum cleaners of data, and they tend to hallucinate. So you have to have a lot of checks and balances on this. But I understand that SAP is doing that here, so that's a positive."

Adoption will be the key to success

The AI-focused tools weren't surprising, as enterprise vendors are all making AI announcements now. But they are good, as code generation is one of generative AI's top use cases, said Jon Reed, co-founder of Diginomica, an enterprise industry analysis firm.

"You can't really argue against [what SAP is doing here], although there's some ongoing debate about how much that changes developers' productivity," Reed said. "This will be sorted out over time, and enterprise development requires more guardrails that challenge some of [those productivity gains]."

With all these different components, having an intelligent agent sitting on your shoulder and pointing you in the right direction is good.

Ultimately, the value of the new tools remains to be seen since adoption is paramount and developers have a lot of choices for tools, he said. SAP must show that these tools can be useful for enterprise developers.

"That has yet to be proven, and the tools can be great. But in the history of this space, we have a lot of tools that looked great that weren't adopted as much as vendors were hoping," Reed said.

The vector database is important. Without that, there's not a good way to bring enterprise-specific information and real-time updates into an LLM, he said.

SAP's approach with the vector database is coherent and makes sense, but companies also have a lot of choices in how they approach AI development, Reed said. SAP needs to convince customers that this is the best approach for them because it leverages their existing SAP investments and all the data in their SAP systems.

"All the announcements are on track and sound good. But it remains to be seen what the adoption level is and if customers like these tools," he said. "Also, some of the pricing is a work in progress. They'll have more coherent answers to that, but it's not detailed yet."

Jim O'Donnell is a TechTarget senior news writer who covers ERP and other enterprise applications for TechTarget Editorial.